THE NTUA SIMULATORS FOR SPACE ROBOTS ON ORBIT Georgios Rekleitis (1) , Ioannis Tortopidis, Iosif Paraskevas, Dimitrios Psarros, Ioannis Kaliakatsos, Ioannis Roditis, and Evangelos Papadopoulos (2) Department of Mechanical Engineering, National Technical University of Athens, Athens 15780, Greece (1) georek@central.ntua.gr , (2) egpapado @central.ntua.gr ABSTRACT The importance of simulators as testbeds for proposed space robotic systems is unquestionable in our days. In this paper, the NTUA approach on both software and hardware space robots simulators, is presented. The software simulator is fully parameterized in order to be capable to simulate any system that consists of a base and a number of serial appendages. The hardware simulator emulates 2D motion in zero gravity, and consists of a two-manipulator space robot moving on top of a granite table. The NTUA simulators are currently at the final stages of their development. 1. INTRODUCTION The commercialization of space introduces the need for robotic devices that can assist humans in the construction, repair, protection and maintenance of space stations or satellites. Such robotic systems are very expensive to build and even more to put successfully on orbit. Furthermore, any error, besides the possible loss of an expensive system, can threaten humans, such as astronauts that may be working with the robot. It is obvious that such systems must be designed and tested to be foolproof before they are tried in their remote, operating environment. Since real experimental tests are practically out of the question, a simulator is the only remaining solution. Simulators in general can be of two kinds. These are software and hardware simulators. Software simulators have the benefit of allowing us to do almost anything we want. The problem is that reality has so many unknown and unmodeled factors, that at best we can approximate reality. Nevertheless, an explicit model of the dynamics of the system can result in a simulation very close to reality. For that reason, several software simulators have been developed over the years, in order to test proposed systems and motion strategies, as in [1]. The main problem encountered while designing a hardware space simulator, is the existence of gravity. Several methods to simulate zero gravity have been proposed. One such approach uses water tanks and systems neutrally buoyant, [2]. The advantage of this approach is that it simulates space motion in 3D, but the existence of the water resistance hampers the realism of the simulation, thus making this approach better for training astronauts in zero gravity slow motions. Another approach is the use of a mechanism that supports the tested robotic system, negating the gravitational force, [3], [4]. This approach, although promising, has yet to overcome the problems due to manipulator singularities, that do not allow any desired motion. Other approaches include throwing systems in deep wells, or parabolic flights, but those are severely limited by the time available for experiments. Yet another approach is a planar simulator, based on practically frictionless motion of the simulated robotic system on a horizontal plane. This motion can be achieved by several methods, such as the use of air bearings [5, 6]. This can be as realistic as close to the actual frictionless motion we can get, but has the obvious disadvantage of the 2D motion restriction. In this paper, the simulators developed at the National Technical University of Athens (NTUA), are presented. These include a software simulator with an animated graphic representation and a planar hardware emulator, based on the frictionless motion of a robot on a granite table, by means of air-bearings. These systems are described in some detail next. 2. SOFTWARE SIMULATOR The software simulator for space robots on orbit consists of three basic components. These are (a) the dynamic equations of motion, (b) the numerical simulation including various control algorithms and, (c) the animated graphic representation. The equations of motion are obtained using the mathematical package Mathematica®, while the system’s behavior under the chosen control method is simulated using a simulation package, such as Simulink®. The animated graphic representation is realized in a program developed in our lab at the NTUA, and is based on OpenGL libraries. This program uses the Simulink® model data output, and produces the desired animation. 2.1. Dynamic Analysis An orbital robotic system consists of a base with several appendages, such as manipulators, communication antennae etc. The motion of any of these appendages changes the geometry of the system, thus influencing its